When Detection Teaches Error: The Case for Critical Artificial Intelligence Literacy in Language Education

Abstract

integrity policies in which automated detection plays a central part, even where the course instructor makes the final judgement. This editorial argues that such policies can lead to unfavorable effects on language learning. Drawing on my experience teaching English for academic purposes and English as a second language at the post-secondary level, I describe an emerging practice, deliberate error injection, in which students insert spelling, grammar, and syntax errors into machine-generated text to lower detection scores. I frame the practice within the postplagiarism era, in which hybrid human–AI writing is increasingly ordinary, and within evidence that detection tools are unreliable and biased against non-native writers, consider how the practice may spread through peer networks, and outline why it matters for learners who are still stabilizing their second language. I propose critical artificial intelligence literacy, integrated into language curricula and supported by process-based assessment, as a more defensible response than prohibition or unguided permission, and I identify the research needed to test these claims.